The price of oil : how does it affect the Royal Dutch Shell stock price?

(1)

The price of oil: How does it affect the

Royal Dutch Shell stock price?

Bachelor Thesis Economics & Finance

Dennis Herkelman 10589511

Supervisor: G. Vala Elias Pimentel Oliveira June 27, 2016

(2)

Statement of Originality

This document is written by student Dennis Herkelman who declares to

take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is

original and that no sources other than those mentioned in the text and

its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the

supervision of completion of the work, not for the contents.

(3)

3. Methodology………...………..…….…….………….…...9 3.1 Data..………...………..……….….…...…..10 3.2 Model……….……...…..10 4. Results………..………...…..……11 5. Conclusion………...………....…….13 6. References………...…....……..14 7. Appendix...………...15 Section A………...15 Section B………...16 Section C………...20

(4)

1. Introduction

This thesis will focus on the adjusted returns of the Royal Dutch Shell stock and the crude Brent oil price between 2006 and 2015. The data will be separated in multiple time periods to compare the effects during increasing and decreasing oil price shocks. I will test if the returns are correlated and if there exists asymmetry. Asymmetry is the phenomenon whereby prices tend to adjust differently according to their direction. I will use a vector autoregression (VAR) model for this research. With this econometric method the linear interdependencies between multiple time series can be tested. Also the lag structure of the variables will be explained in this model. Previous studies concerning this topic show different effects of the oil price on the stock market.

Driesprong, Jacobsen and Maat (2007) find an overall significant negative correlation while Kilian and Park (2009) contradict these findings by observing a positive correlation. Even when focussing specific on oil related companies the results are not clear. The research of Lanza, Manera, Grasso and Giovanni (2005) about the effect of the oil price among different oil companies results in both positive and negative correlations.

Shell stocks are traded on different markets worldwide, including the NYSE. Knowing the correlation between the oil price and the Shell stock price is important for creating investment portfolios. Most portfolios consist of multiple large companies including Shell. When combining stocks with commodities in a portfolio it is useful to know the effects on each other. Investors want to make efficient portfolios by using a Sharpe ratio for example (Bodie, Kane and Marcus, 2014). Therefore the right understanding of the relationship is important. Also for hedging with

commodities for the company itself, this information is crucial. Companies like Shell can offset potential losses and gains by developing forward contracts. Besides the importance of creating a portfolio or hedging with commodities, the forecasting aspect is important. The effects of an oil price change on a Shell stock gives investors the opportunity to anticipate. For example when OPEC makes decisions about future production or setting prices it is useful to know what the effects on Shell will be.

In the recent decade relatively large oil price changes occurred. In this decade there were periods with large price increases and also price decreases. According to the U.S. Energy Information Administration the spot price of a barrel crude Brent oil dropped between 2008 and 2009. The price decreased with about hundred dollar per barrel, which is more than a 70% price decrease. The period after that consisted of a price growth until 2011. The price of one barrel increased more than eighty dollar, which is more than a 200%. The following four years were stable, the oil price fluctuated between a twenty dollar bandwidth. Oil is still one of the most important commodities these days. Price changes affect the financial markets worldwide. In 2013 the total oil consumption rose above the 90 million barrels a day.

(5)

Royal Dutch Shell is one of the seven major oil related companies of the world. The Dutch-British company is active and fully integrated in the total production of oil. Shell’s revenues in 2014 summed up to 421 billion dollars in more than 140 countries according to their annual report. These statistics are comparable with the American Exxon Mobil (394 billion) and the British BP (354 billion). These three oil companies perform in a category below the largest oil producers from China and Saudi Arabia. Shell’s upstream and downstream activities in the oil business are

associated with Brent oil. The Brent oilfields, located in the North Sea between the United Kingdom and Norway, are fully controlled and operated by Shell. Brent oil is together with the West Texas Intermediate (WTI) and Dubai Crude one of the three benchmark oil brands that serve as a reference. Together with their large revenues and distribution they are in possession of an inventory crude oil. The complete inventory of Shell in 2014 consisted of 6,567 million barrels (Shell, 2015). Followed by Exxon Mobil with a reserve of 6,048 million barrels (Exxon Mobil, 2015) and BP witha reserve of 4,961 million barrels (BP, 2015). Taken these amounts in consideration the price of oil matters for a company’s performance and liquidity.

The remainder of this thesis will be structured as follows. Section 2 gives an overview of the existing literature done so far on the related topic. The existing dynamics of the stock and oil market will be elaborated. Section 3 describes the data and gives an explanation of the VAR model. In section 4 the empirical results of the tests will be discussed. Finally, section 5 will provide a summary of the results and a conclusion of this paper.

2. Literature review

In this chapter the previous research about the related topic will be discussed. The aim of this part is to formulate a mechanism on the basis of other research results. First the historical behaviour and importance of the oil price for the U.S. GDP will be discussed. Subsequently the dynamics of the gasoline market are examined. Thirdly the U.S. stock market and the oil related companies will be considered. Combining this information will create a clear view of the multiple effects

regarding the Shell stock. At the end of this literature review I will describe a recapitulation of my presumable expectations according to the existing dynamics.

2.1 U.S. GDP

The oil price effects on the economic growth is an important mechanism for Shell stocks. The U.S. GDP is taken as reference because it is one of the largest economies in the world. Additionally, the U.S. GDP is relevant due to the fact that the NYSE index is included. Also the stock value of Shell is taken from the U.S. stock market.

Since the WWII oil became a more important part in explaining and predicting the financial markets and growth of economies (Hamilton, 1983). Hamilton’s research find empirical evidence

(6)

that there exists a correlation between the oil price and the real output in the U.S.. His results show that the oil price rose drastically prior to seven out of the eight financial recessions in the U.S.. Hamilton states that the oil price has significant effect on the economic performance but it is not the decisive factor of a recession. However he discovers a significant negative correlation between the U.S. GDP and the oil price, which is also asymmetric. He claims that increasing oil prices retard the economic growth more than falling prices stimulate. The degree of predicting the economic performance on the basis of the oil price decreases over time according to Hooker (1996). The importance of the oil price after 1973 has changed he explains. The increasing oil price shocks seems to retain its negative correlation on the economy, but smaller. The decreasing oil prices on the other hand do not show any significant correlations with the economy since 1983. According to Balke, Brown and Yücel (2002) there is a negative and asymmetric correlation between oil prices and the U.S. (GDP). They tested the asymmetric effects of oil price changes in their paper. A VAR model is used with the natural logarithms of the GDP, commodities index, oil price and the bond rates with different maturities. More expensive oil prices cause higher

transportation, production, heating cost, which can stagger the company’s earnings. The reason they explain for this asymmetry is because increasing oil prices provide financial distress and necessary adjustment costs. When oil prices decrease the financial stress and adjustment costs do not occur. Despite the effect of oil on the economic growth it also works in the opposite direction. This means that the economic growths also affects the oil price. Kilian and Park (2009) describe that when the markets are growing and the production rises, the demand for oil increases. This phenomenon is explained by the supply and demand equilibrium that occurs on efficient and competitive markets. According to this theory it counter effects the negative correlation.

Eventually both studies show a small but significant negative correlation between the GDP and the oil price.

2.2 Gasoline market

When explaining a possible mechanism behind the oil price effects on a Shell stock, the gasoline market is a good reference. About 76% of the total oil products that were consumed in the U.S. are gasoline, including diesel and jet fuels. Since Shell is active in the downstream with more than 44,000 petrol stations worldwide the gasoline market matters.

According to Bettendorf, van der Geest and Varkevisser (2003) consumer gasoline prices react significant asymmetric to oil price changes on Monday, Thursday and Friday. In their research they used an error correction model (ECM) with the data from the Dutch gasoline prices (Euro 95) between 1996 and 2001 and compared those with the oil price (brand not mentioned). These empirical findings are partly due to the fact that there exists low competition between gasoline retailers. This low competition occurs because of vertical agreements they explain. Oil companies

(7)

like Shell, Exxon Mobil and BP are active in the whole chain of oil production. Starting from the beginning, when oil is retrieved from the oilrigs, until the end by selling the refined products. When the gas stations decide to lower their prices, they are aware that the competitors will follow immediately. In this case the stations have no incentive to cut the prices because it will lead to a lower profit. Another explanation by Bettendorf et al (2003) is that even the smaller independent retailers get their gasoline from the big oil providing companies, which Shell is part of. They describe that the oil retailers are committed to long term supply contracts and the big oil companies lead that market. Borenstein, Cameron and Gilbert (1992) also support the market power of some producers and distributors selling gasoline and diesel fuel. In their paper they test the time of the price adjustments of gasoline according to the oil price changes with a simple linear model. Followed by an ECM including the different lags. They also find results that affirm the theory that the gasoline prices react slower and less heavy to oil price decreases. The results show that the decreasing price adjustments take twice as much the time as the price increase adjustment. This is because the oil producers need to compensate for the decreased revenues for oil (also for gasoline). Huntington (1998) found that regular consumers, meaning personal transport, respond symmetrically to gasoline prices.

According to this information conclusion can be made about the performance of Shell, affecting their stock price. Because of Shell’s large distribution network the stock value is partially affected by the gasoline market. Increasing gasoline prices due to higher oil prices provide higher profits, which is beneficial for Shell, but it decreases the demand. Besides that, the gasoline that is sold to other retailers has a fixed price. The price changes could be unfavourable in this case. When gasoline prices drop eventually, from an oil price decrease, the profits drop. The period between the oil price decrease and the gasoline price decrease is for compensating the losses due to lower profits. In this situation the fixed gasoline prices for the retailers are favourable. Both increasing and decreasing prices include advantages and disadvantages for Shell. The actions of the company are made to compensate for potential losses. Changes of the gasoline prices do not have a large effect on the performance this way but the correlation is presumably positive.

2.3 U.S. Stock market

Besides the economic growth, the stock market is influenced by the oil price according to multiple studies. Again, the U.S. stock market is used because it is relevant for the Shell stock and the NYSE index.

Driesprong et al. (2007) did research on the correlation between the worldwide markets, including the U.S., and the oil price between 1973 and 2003. Crude Brent oil, WTI and Dubai Crude are used as benchmark prices. Their results show a significant negative correlation between the stock and oil values. Stronger and more significant effects are found in the markets of the developed

(8)

countries when compared with the emerging markets. The empirical results show that the stock market reacts with a 6 day lag to the oil price. This means that the oil price changes have a delayed effect on the stock market. Investors take time to adjust the effects of oil prices on the stock values. According to Kilian and Park (2009) there exists a positive correlation between the stock returns and the crude oil price. Kilian and Park’s research about the asymmetric effects of the crude oil price and the U.S. stock market. In their research a VAR model is used. They explain that unanticipated global economic expansion drives the oil price up. In this case the stock market returns outperform the drag of the oil price on the economy.

Pescatori (2008) claims that no stable correlation exists at all. His results show mostly small negative insignificant correlation. He used the S&P500 together with crude oil (WTI) to create a scatterplot of the weekly growth of the market and oil between 1998 and 2008. The results were slightly negative (-0.021) and insignificant. However he does not exclude the possibility that there is a significant negative relation when using daily or monthly periods, different market or another time series. The only significant negative correlation was found in the transportation sector of the Dow Jones. This could be explained by their dependence of fuel which results in a direct share price effect.

Looking at the research Miller and Ratti (2009) did with a vector error correction (VEC) model on the long-run effects of the oil price on the stock market gives a broader view. Their research starts in 1971 until 2008 and studies multiple separate time periods. The correlation between the crude oil price and the stock returns decreases over time they found. Not only the correlation became less, also the results became more insignificant. In other words the oil price gets less influence on the stock market. Another paper which examined the relation between the oil price and stock returns is done by Sadorsky (1999). A VAR model was used with the monthly natural logarithm of the U.S. industrial production, S&P500 and the oil price between 1947 and 1996. He claims that the oil price has significant effect on the stock returns but stock returns have no effect on the oil price, which contradicts the results of Kilian and Park (2009). His empirical research also shows a swift in the dependence of oil since 1986 like Miller and Ratti (2009) found. Sadorsky also tested for asymmetry between the oil price shocks and the stock returns where no evidence was found.

2.4 Oil related companies

Companies that are specific related to oil have different reactions to oil price changes than other companies. Studies on the oil related companies are the best comparison for this research. This section will discuss the findings about the different effects of the oil specific industry. This includes the transportation market and the companies which focus on the upstream and downstream of oil production. Also the lag structure for the oil related companies will be appointed.

(9)

Huang, Masulis and Stoll (1995) did a partly similar research on this subject. Their aim was to find out about the effects of the future oil prices on stock markets and the separate oil related

companies Exxon, Mobil and Chevron. Also transportation companies are taken into

consideration. According to Huang et al. (1995) the future oil prices are reflected first to the oil dependant markets and transmit with a lag to the regular stocks. Notable is that the results conclude that the oil price only affects oil producing companies. The transportation market was unaffected as well as the regular stock market. Driesprong et al. (2007) also support the delayed reaction of the oil price on the regular stock market by a 6-day lag. Surprisingly, this effect does not occur when predicting the oil related companies stock. These findings could be explained by the investors who reflect the oil prices immediately to the oil dependant firms. Both Driesprong and Huang find an immediate effect on the oil specific industry companies. The oil stock returns show a significant positive relation with the oil price with a one day lag. The methodology Huang et. al used for their findings will be taken as guideline for this research.

Lanza et al. (2005) did specific research to the oil price effects on oil related companies. In their VEC model Royal Dutch Shell, BP, Chevron-Texaco, Exxon-Mobil, Eni and Total-Fina-Elf are included. The fully integrated oil companies show different results. For example Chevron, Exxon and Eni have a positive correlation with the oil price while Shell, Total and BP have a negative correlation. The dependence of oil is partly due to the degree of integration. They explain that the companies which focus only on the downstream could be penalized by increasing oil prices since they are unable to transfer the costs into the refined products. On the other hand upstream

companies benefit from this situation.

Taken all the previous studies into consideration I expect a small positive correlation between a Shell stock and the oil prices. The companies that are active on the gasoline market recover themselves by the adjustment speed of the fuel prices. Which will have therefore a small effect on the firm’s value. Even though different results are found I expect for Shell a small lagged positive correlation with the oil price. Since oil prices are transferred into the oil sector of the market first, the lag should not be larger than one day. Also due to the fact that oil price decreases affect the economic growth more than price increases stimulate I expect asymmetry. This asymmetry would appear by a larger effect of an oil price decrease.

3 Methodology

This section will explain the methodology used to research correlation and asymmetry. First the data will be discussed. Subsequently the VAR model will be explained together with the tests that are used to determine an answer to the research question.

(10)

3.2 Data

The data of the adjusted close price of a Shell stock from the New York Stock Exchange (NYSE) will be used. The adjusted close price includes any other gains from possessing a stock, for example dividends. This prices gives a better view of the total returns. According to the previous studies efficient markets are assumed and the oil price changes will be reflected on the (oil-) companies within days. For this reason the daily returns of the oil price, NYSE index and Shell stock are taken. The data of the NYSE will be used in the model as control variable since Shell is probably dependent on the overall performance of the market. For the daily oil price I took the values of a barrel crude Brent oil. The time period for the data starts from 2006 until 2015. The reason for this time period could be explained on the basis of the fluctuations. This time window consists of a decreasing oil price shock between 03/07/2008 and 11/05/2009. After this period an increasing oil price shock occurred between 08/02/2010 and 02/05/2011. Followed by this grow and fall, a stable period on the oil price appeared between 08/08/2012 and 16/07/2014. To do proper research to this question it is important to take both shock and a stable period into consideration. A conclusion about asymmetry could be made when comparing the results of the increasing and decreasing oil price period. Also the full period and the stable period are useful to examine the overall correlation between the Shell stock and the oil price.

3.3 Model

To investigate the correlation and asymmetry over time a vector auto regression (VAR) model will be used. To test the effects on the Shellstock returns I use the first difference of the variables. The first difference of the Shell stock will be specified as: (𝑆ℎ𝑒𝑙𝑙 𝑡) − (𝑆ℎ𝑒𝑙𝑙 𝑡−1), the first difference of the oil price as: (𝑂𝑖𝑙 𝑡) − (𝑂𝑖𝑙 𝑡−1), and the first difference of the NYSE index as: (𝑁𝑌𝑆𝐸 𝑡) − (𝑁𝑌𝑆𝐸 𝑡−1). All the test will be done for each time series. This means that the increasing oil price period, decreasing oil price period, stable oil price period and full period will be researched separately. Primarily the first differences need to be tested on stationarity for all time periods. The assumption that no autocorrelation exist in neither the first differences or the residuals need to satisfy. After that the appropriate lag structure for the four time series will be determined. According to Franses, van Dijk and Opschoor (2014) seasonality adjustments need to be made mostly when time series are observed monthly or quarterly. Also they suggest that these adjustments need to be made if the time series display a certain region. For example the

consumption of oil could be higher during cold months. Personal transport by cars and heating will be higher during winter months when compared with summer months. As a result seasonal

adjustment is effective when seasonal patterns exist in a time series. Since the market of both oil and Shell are worldwide, it is not sensitive for these season variations. Seasonal adjustment will be unnecessary in this case.

(11)

The correlation and asymmetry can be tested when the model satisfies the assumption for all the four separate time periods. The Cholesky decomposition will be used to find the specific short-term effect of the oil shocks. When using this impulse response function the particular effect of the oil shock can be measured on the basis of the chosen lag order. Hereafter the explanation of the variance of the variables on each other will be determined with the variance decomposition. To determine the long-term relation between the oil shocks and the Shell stock the Johansen long run model will be used. Finally according to the Granger causality test the degree of forecasting will be examined. The Granger causality can measure the ability of predicting future values of a time series by using the previous values of another time series.

4 Results

The first test for all periods is the stationary test to find any autocorrelation. The returns of oil, Shell and the NYSE in all the time series are stationary according to the unit root test. In Appendix section B the results are displayed in Table 2. To test autocorrelation between the residuals of the time series the LM autocorrelation in Eviews is used. No significant autocorrelation is found in the four time periods. The results of the residuals autocorrelation test can be observed in Table 3 in the Appendix section B. After the assumption that no autocorrelation exists, in neither the returns or their residuals, the lag criteria can be examined. The Eviews outputs in section C of the Appendix will show the lag criteria for each period. Remarkable is that the full period, the fall period and the grow period have a 1-day lag order according to the most criteria. However the stable period results in a no lag criteria. In other words the oil price changes have no delayed reaction on the stock.

To test the correlation and asymmetry the Cholesky decomposition is used. With this test it is possible to separate the different shocks that affect the Shell stock. The effect of the oil price on a Shell stock is what I want to measure. This means that the 1-day lag effect of the values of the Shell stock and the NYSE will be set to zero. Now only the specific 1-day lag effect of the oil price change on the Shell stock can be observed. The results of this decomposition are explained in Table 4 in section B of the Appendix. The graphs of the Cholesky decomposition that show only the effect of the oil price on the Shell stock are shown in Figure 2 in section B of the Appendix. Both Table 4 and Figure 2 directly show a positive relation between the oil price and Shell independent of the behaviour of the oil price. All series have a significant effect which supports most of the earlier papers that are discussed. What we also can read is that the effects on a Shell stock in the period of a decreasing oil price is approximately twice as large as in the grow period. This result proves large asymmetry depending on the direction of the shock. A remarkable results is the low effect during a stable oil price period. Meaning that the oil price changes have a significant lower effect if no large shocks occur. This small correlation could be reflected by the

(12)

previous finding of the lag criteria result. Another explanation could be the decreasing effect of the oil price on stocks over time. This would be unlikely because these periods are relatively short and follow each other immediately.

Besides the asymmetry of the coefficients I tested the contribution of the variances of the

variables. Table 5 in the Appendix section B shows the variance of the return of a Shell stock that is explained by the variance of the return of the oil price. According to this results it is clear that the negative and positive oil shocks have no asymmetry in their variance. Both fall and grow periods show that the variance of Shell is about 16% due to the variance of the oil price. Again the stable oil price period results in a different outcome. Only nearly 2% of a Shell stock’s volatility is caused by the volatility of the oil price. This means that during oil shocks the variances are more correlation with each other when compared with a stable period.

The Johansen long-run model is used to examine the accumulated effects of the NSYE and oil price on the Shell stock. It is clear that the long-run coefficients of the oil price are positive over all the time series according to Table 6 of section B of the Appendix. The NYSE shows a negative correlation with the return of Shell. These results were predicted in advance. Additionally, asymmetry occurs among the different time periods. The asymmetry is in the opposite direction when compared with the Cholesky decomposition. In this case the total effect of the increasing oil price shocks is 10 times larger than the decreasing oil price shocks. Meaning that the effects of the grow period have no large contemporaneous effect on the stock price, like the fall period, but the overall effect is larger. This phenomenon could be explained by the quick responses of investors by anticipating on oil shocks and the slow response of changing economic conditions.

Finally I used the Granger causality test to examine the causality of the variables. When the predictions of the Shell returns based on its own past values and the past values of another variable are better than prediction based on only the past values of the Shell returns, the Shell returns are Granger caused by the other variable. Each period is tested on a 1, 2, 4 and 8-day lag structure. According to the results displayed in Table 7 in section B of the Appendix it is clear that only in the full period the level of causality is significant and decreasing with the number of lags. The separate periods of falling, growing and stable oil prices have no sign of causality. The reliability could be questionable because the separate periods consist of a smaller sample compared with the full period. There could be a biased based on the sample size affecting the standard error. When comparing these results, the causality of the NYSE returns on the oil price is standing out. For the full period and the fall period the causality was significant for the 1, 2, 4 and 8-day lag orders. Also these causalities decrease with the amount of lags. This refers to previous papers about the effects of the, in this case, U.S. stock market on the oil price. In the fall period the NYSE is a good indicator of how the oil price will react in contrast with the grow and stable period. This result

(13)

points at the supply and demand function whereby the oil price is affected and predictable by the stock market performance.

5. Conclusion

The research question of this thesis is how the Royal Dutch Shell stock price reacts on crude Brent oil price shocks during multiple time periods. The results of the VAR model support the main observation from the earlier studies. The correlation between the oil shocks and the returns of the stock are positive in all situations, as expected. However there exists asymmetry in the returns when comparing increasing and decreasing oil price shocks. When a shock with decreasing oil prices occurs, the returns of the Shell stock react twice as large compared with an increasing oil price shock. Besides this asymmetry the correlation between the stock and the oil price during a stable period is very low. Meaning that the oil price changes affect the stock for a small part. This phenomenon is also reflected in the variance explanation. In stable oil price periods the effect on a Shell stock is much lower. In both increasing and decreasing shocks the variance is responsible for approximately the same proportion of the stock price variance. Which means there is no

asymmetry. With the long-run model the total effect of an oil price change on the stock is

measured. Also in this model asymmetry is found, only in opposite direction. During increasing oil price shocks the immediate effect on the stock is smaller but the accumulated effect is larger. The decreasing oil price shocks have a quick response on the stock but has only a small influence in the long run. Furthermore the causality tests give no significant results about the prediction during oil price shocks. In the end the conclusion can be made that the results support most of the existing literature. The correlation between the oil price and Shell is positive and subject to asymmetry.

(14)

6. References:

Balke, N. S., Brown, S. P. A., & Yücel, M. K. (2002). Oil Price Shocks and the U.S. Economy: Where Does the Asymmetry Originate? The Energy Journal, Vol. 23, No. 3 (2002), pp. 27-52 Bettendorf, L., van der Geest, S. A., & Varkevisser, M. (2003). Price asymmetry in the Dutch retail gasoline market. Energy Economics 25 (2003), pp. 669-689

Bodie, Z., Kane, A., & Marcus, A. (2014). Investments (10th Global Edition). United Kingdom: McGraw-Hill Education

Borenstein, S., Cameron, A. C., & Gilbert, R. (1992). Do gasoline prices respond asymmetrically to crude oil price changes? NBER working paper series, working paper No. 4138

BP (2015). Annual Report 2015. Retrieved from http://www.bp.com/

Driesprong, G., Jacobsen, B., & Maat, B. (2007). Striking oil: Another puzzle? Journal o Financial Economics 89 (2008), pp. 307-327

Exxon Mobil (2015). 2015 Financial Statements. Retrieved from http://corporate.exxonmobil.com/

Franses, P. H., van Dijk, D., & Opschoor, A. (2014). Time Series Models for Business and Economic Forecasting (second edition). United Kingdom: Cambridge University Press

Hamilton, J. D. (1983). Oil and the Macroeconomy since World War II. Journal of Political Economy, Vol. 91, No. 2 (Apr., 1983), pp. 228-248

Hooker, M. A. (1996). What happened to the oil price-macroeconomy relationship? Journal of Monetary Economics 38 (1996), pp. 195-213

Huang, R. D., Masulis, R. W., & Stoll, H. R. (1995). Energy Shocks and Financial Markets. Journal of Futures Markets 16(1), February 1996

Huntington, G. (1998). Crude Oil Prices and U.S. Economic Performance: where Does the Asymmetry Reside? The Energy Journal 19(4), pp. 107-132

Kilian, L., & Park, C. (2009). The impact of oil price shocks on the U.S. stock market. International Economic Review, Vol. 50, No. 4, November 2009

Lanza, A., Manera, M., Grasso, M., & Giovanni, M. (2005). Long-run models of oil stock prices. Environmental Modelling & Software 20 (2005), pp. 1423-1430

Miller, J. I., & Ratti, R. A. (2009). Crude oil and stock markets: Stability, instability, and bubbles. Energy Economics 31 (2009), pp. 559-568

Pescatori, A., & Mowry, B. (2008). Do Oil Prices Directly Affect the Stock Market? Economic Trends

Sadorsky, P. (1999). Oil price shocks and stock market activity. Energy Economics 21 (2003), pp. 449-469

Shell (2015). Royal Dutch Shell annual report 2015. Retrieved from http://www.shell.nl/

U.S. Energy Information Administration. Petrolium & Other Liquids. Retrieved from

(15)

7. Appendix

Section A

Summary statistics

Table 1 Descriptive statistics

Note: All the daily values are expressed in USD

Figure 1 Daily oil and stock price

Note: All the daily values are expressed in USD

Descriptive stats Adj Close RDS Barrel Close NYSE

Mean 49.86295 85.80617 8601.387 Median 51.18000 84.21000 8442.985 Maximum 74.49000 143.9500 11239.66 Minimum 26.31000 33.73000 4226.310 Std. Dev. 9.694679 24.66555 1540.169 Skewness 0.126752 -0.089064 -0.280328 Kurtosis 2.610767 1.810753 2.492992 Observations 2506 2506 2506

(16)

Section B

Table and figure overview

Table 2 Variable stationarity

Variable/period Full Fall Grow Stable

RDS -52.02866*** -13.79299*** -17.00716*** -23.18990***

Oil -48.12783*** -13.91186*** -18.67751*** -21.45152***

NYSE -53.80980*** -13.15426*** -18.33201*** -22.58780***

The null hypothesis suggest that the variable has a unit root. The table shows the t-statistics of the Dickey-Fuller test

*** = reject the null hypothesis at the 1% significance level ** = reject the null hypothesis at the 5% significance level * = reject the null hypothesis at the 10% significance level

Table 3 VAR residual stationarity

The null hypothesis suggest no serial correlation between the residuals. *** = reject the null hypothesis at the 1% significance level

** = reject the null hypothesis at the 5% significance level * = reject the null hypothesis at the 10% significance level

Autocorrelation test/period Full Fall Grow Stable

(17)

Table 4 Cholesky decomposition

Response/period Full Fall Grow Stable

RDS to oil (1-day lag)

0.278*** 0.535*** 0.264*** 0.074***

*** = statistically significant at the 1% level ** = statistically significant at the 5% level * = statistically significant at the 10% level

(18)

Table 5 Variance decomposition

The percentages show the proportion of the Shell stock variance that is explained by oil price variance

Figure 3 Variance decomposition

Variance/period Full Fall Grow Stable

(19)

Table 6 Johansen long-run model

The model refers to the long-run coefficients of the first difference of the variables on the first difference of the Shell stock for each period

*** = statistically significant at the 1% level ** = statistically significant at the 5% level * = statistically significant at the 10% level

Table 7 Granger causality

The null hypothesis suggest no Granger causality. The table shows the F-statistics of the Granger causality test

*** = reject the null hypothesis at the 1% significance level ** = reject the null hypothesis at the 5% significance level * = reject the null hypothesis at the 10% significance level

Constant D(Barrel) D(NYSE)

Full period 0.025255 0.899511*** -0.014794***

Fall period -0.022696 0.212145*** -0.002311***

Grow period -0.127781 2.165660*** -0.034782***

Stable period 0.153642 1.153018*** -0.016559***

Causality/period Full Fall Grow Stable

RDS by oil 1-day lag 8.74856*** 0.20484 3.30618* 0.70262

RDS by oil 2-day lag 5.16193*** 0.49777 1.68693 0.36520

RDS by oil 4-day lag 2.86764** 0.36990 1.04976 0.61105

RDS by oil 8-day lag 1.70258* 0.47360 0.75090 0.80899

Causality/period Full Fall Grow Stable

Oil by NYSE 1-day lag 51.5721*** 16.7101*** 4.97217** 1.94092 Oil by NYSE 2-day lag 26.0118*** 9.31745*** 2.94747* 1.05864 Oil by NYSE 4-day lag 13.2739*** 5.41776*** 1.42353 1.72601 Oil by NYSE 8-day lag 6.70405*** 2.75331*** 1.25580 1.38071

(20)

Section C Eviews outputs

Lag order criteria full period

Lag order criteria grow period

VAR Lag Order Selection Criteria

Endogenous variables: D(ADJ_CLOSE_RDS) D(BARREL) D(CLOSE_NYSE)

Exogenous variables: C Date: 06/08/16 Time: 12:37 Sample: 2/08/2010 5/02/2011 Included observations: 300

Lag LogL LR FPE AIC SC HQ

0 -2418.790 NA 2062.422 16.14527 16.18231* 16.16009* 1 -2406.717 23.82395* 2020.610* 16.12478* 16.27293 16.18407 2 -2402.696 7.855025 2088.845 16.15797 16.41724 16.26173 3 -2396.877 11.24943 2133.724 16.17918 16.54956 16.32741 4 -2392.286 8.783748 2197.570 16.20858 16.69007 16.40127 5 -2387.800 8.493958 2265.023 16.23867 16.83127 16.47583 6 -2382.061 10.75192 2315.272 16.26040 16.96412 16.54203 7 -2375.503 12.15442 2353.935 16.27668 17.09152 16.60278 8 -2371.209 7.870679 2429.872 16.30806 17.23401 16.67863

* indicates lag order selected by the criterion

LR: sequential modified LR test statistic (each test at 5% level) FPE: Final prediction error

AIC: Akaike information criterion SC: Schwarz information criterion HQ: Hannan-Quinn information criterion VAR Lag Order Selection Criteria

0 -21577.07 NA 6535.596 17.29865 17.30565 17.30119 1 -21513.84 126.2396 6257.618* 17.25519* 17.28319* 17.26535* 2 -21506.88 13.88122 6267.859 17.25682 17.30582 17.27461 3 -21502.06 9.606420 6288.884 17.26017 17.33017 17.28559 4 -21497.43 9.219151 6310.934 17.26367 17.35468 17.29671 5 -21492.09 10.61031 6329.482 17.26660 17.37861 17.30727 6 -21488.66 6.803432 6357.819 17.27107 17.40408 17.31936 7 -21486.42 4.435237 6392.381 17.27649 17.43050 17.33241 8 -21482.34 8.090073 6417.615 17.28043 17.45544 17.34397 9 -21471.72 20.98760 6409.328 17.27914 17.47515 17.35031 10 -21462.73 17.77208* 6409.346 17.27914 17.49615 17.35793

AIC: Akaike information criterion SC: Schwarz information criterion HQ: Hannan-Quinn information criterion

(21)

Lag order criteria fall period

Endogenous variables: D(_F_ADJ_CLOSE_RDS) D(_F_BARREL) D(_F_CLOSE_NYSE) Exogenous variables: C

Date: 06/08/16 Time: 12:08 Sample: 7/03/2008 5/11/2009 Included observations: 206

0 -2053.502 NA 94127.29 19.96603 20.01450* 19.98563 1 -2035.494 35.31505* 86246.89* 19.87859* 20.07244 19.95699* 2 -2027.860 14.74907 87403.96 19.89185 20.23110 20.02905 3 -2026.497 2.593991 94144.91 19.96599 20.45063 20.16200 4 -2024.180 4.341612 100483.5 20.03088 20.66091 20.28568 5 -2020.788 6.257614 106152.2 20.08532 20.86075 20.39893 6 -2016.615 7.576899 111315.8 20.13218 21.05300 20.50459 7 -2011.862 8.489238 116104.2 20.17342 21.23963 20.60463 8 -2010.134 3.036865 124741.3 20.24402 21.45563 20.73404

Lag order criteria stable period

0 -3708.867 NA 1152.573* 15.56338* 15.58959* 15.57369* 1 -3705.646 6.387803 1180.843 15.58761 15.69246 15.62884 2 -3700.613 9.917739 1200.656 15.60425 15.78772 15.67639 3 -3692.820 15.26027 1206.760 15.60931 15.87141 15.71236 4 -3689.734 6.002336 1237.089 15.63411 15.97485 15.76808 5 -3687.460 4.395576 1272.517 15.66231 16.08168 15.82720 6 -3685.820 3.149951 1312.464 15.69317 16.19117 15.88897 7 -3680.781 9.614111 1334.534 15.70977 16.28641 15.93650 8 -3671.376 17.82343* 1332.394 15.70808 16.36335 15.96572

The price of oil : how does it affect the Royal Dutch Shell stock price?